Rhythmicity of Coastal Marine Picoeukaryotes, Bacteria and Archaea Despite Irregular Environmental Perturbations

Home , Bacterioplankton, Cyanobacteria, Picoeukaryote

Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental perturbations Stefan Lambert, Margot Tragin, Jean-Claude Lozano, Jean-François Ghiglione, Daniel Vaulot, François-Yves Bouget, Pierre Galand

To cite this version:

Stefan Lambert, Margot Tragin, Jean-Claude Lozano, Jean-François Ghiglione, Daniel Vaulot, et al.. Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental perturbations. ISME Journal, Nature Publishing Group, 2019, 13 (2), pp.388-401. 10.1038/s41396- 018-0281-z. hal-02326251

HAL Id: hal-02326251 https://hal.archives-ouvertes.fr/hal-02326251 Submitted on 19 Nov 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental perturbations Stefan Lambert, Margot Tragin, Jean-Claude Lozano, Jean-François Ghiglione, Daniel Vaulot, François-Yves Bouget, Pierre Galand

To cite this version:

HAL Id: hal-02326251 https://hal.archives-ouvertes.fr/hal-02326251 Submitted on 19 Nov 2020

1 ● 2 ● 1 ● 1 ● 2 ● Stefan Lambert Margot Tragin Jean-Claude Lozano Jean-François Ghiglione Daniel Vaulot 1 3 François-Yves Bouget ● Pierre E. Galand

Received: 25 May 2018 / Revised: 20 August 2018 / Accepted: 26 August 2018

Abstract Seasonality in marine microorganisms has been classically observed in phytoplankton blooms, and more recently studied at the community level in prokaryotes, but rarely investigated at the scale of individual microbial taxa. Here we test if specific marine eukaryotic phytoplankton, bacterial and archaeal taxa display yearly rhythms at a coastal site impacted by irregular environmental perturbations. Our seven-year study in the Bay of Banyuls (North Western Mediterranean Sea) shows that despite some fluctuating environmental conditions, many microbial taxa displayed significant yearly rhythms. The robust rhythmicity was found in both autotrophs (picoeukaryotes and cyanobacteria) and heterotrophic prokaryotes. Sporadic meteorological events and irregular nutrient supplies did, however, trigger the appearance of less common non-rhythmic 1234567890();,: 1234567890();,: taxa. Among the environmental parameters that were measured, the main drivers of rhythmicity were temperature and day length. Seasonal autotrophs may thus be setting the pace for rhythmic heterotrophs. Similar environmental niches may be driving seasonality as well. The observed strong association between Micromonas and SAR11, which both need thiamine precursors for growth, could be a first indication that shared nutritional niches may explain some rhythmic patterns of co- occurrence.

Introduction migrations in animals, flowering in plants and blooms in plankton communities [1–3]. Phytoplanktonic blooms in Regular and predictable fluctuations of environmental temperate oceanic areas are a typical example of seasonal parameters have a great impact on life. Seasonality sets the events. Several classical theories, from Sverdrup’s “Critical pace for many reoccurring life events, such as mating or Depths Hypothesis” [4] to Behrenfeld’s “Dilution-Recou- pling Hypothesis” [5], have attempted to explain the mechanisms triggering bloom formation. However, these theories do not aim to explain the reoccurrence and sea- Electronic supplementary material The online version of this article sonality of specific microbial taxa. In macroscopic organ- (https://doi.org/10.1038/s41396-018-0281-z) contains supplementary fi material, which is available to authorized users. isms, seasonality results from a ne interplay between external environmental factors and the internal circadian * François-Yves Bouget clock, which is an endogenous timekeeper [6]. In marine [email protected] microorganisms, circadian rhythms are less well known and * Pierre E. Galand they have been reported only in cyanobacteria and in some [email protected] eukaryotic microalgae [7–10]. However, the effect of environmental forcing on the seasonality of entire bacterial 1 ’ CNRS, Laboratoire d Océanographie Microbienne (LOMIC), communities has been studied more extensively and reoc- Observatoire Océanologique de Banyuls, Sorbonne Université, Banyuls sur Mer, Paris, France curring microbial communities are often observed 2 responding to environmental changes [11–15]. CNRS, UMR7144, Station Biologique de Roscoff, Sorbonne fl Université, Roscoff, Paris, France Oceans are uctuating habitats that are often marked by a strong seasonality. These regular environmental changes 3 CNRS, Laboratoire d’Ecogéochimie des Environnements Benthiques (LECOB), Observatoire Océanologique de Banyuls, allow for an overall high microbial community diversity, Sorbonne Université, Banyuls sur Mer, Paris, France since the environment can accommodate different species in S. Lambert et al. the same space, but at different times of the year [16]. factors that could contribute to microbial seasonality. We Within a year, diversity also varies locally with peaks also used statistical tools to quantify the rhythmicity of the observed in winter at high latitudes [15, 17] and community picoplankton and to detect patterns of co-occurrence composition changes with seasons. Seasonal cycles in between eukaryotic picophytoplankton (less than 3 µm), abiotic and/or biotic factors drive these community changes bacteria and archaea. [18, 19]. To understand the seasonality of marine microbial communities, several long term sampling sites have been established within the last couple decades leading to some Materials and methods important findings on the seasonality of major microbial groups in the surface of the ocean [14, 20–23] and the Environmental sampling reoccurring patterns of microbial community composition [12, 24]. Surface seawater (3 m depth) was collected roughly every Most of earlier studies focused on bacteria and there are 2 weeks from October 2007 to January 2015 at the Service only few reports on the seasonality of the other domains of d’Observation du Laboratoire Arago (SOLA) sampling life. For marine archaea, it has been shown that both rare station (42°31′N, 03°11′E) in the Bay of Banyuls-sur-Mer, and abundant members of the community were re-occurring North Western Mediterranean Sea, France. Seawater was seasonally and that different ecotypes of archaea had dif- collected in 10 l Niskin bottles and then kept in 10 l carboys ferent seasonal patterns [20, 25]. For phytoplankton, evi- until arrival to the laboratory within one hour. A subsample dence for global patterns of temporal dynamics were of 5 l was prefiltered through 3 μm pore-size polycarbonate obtained by compiling seasonal data of chlorophyll a con- filters (Merck-Millipore, Darmstadt, Germany), and the centrations [26]. Molecular techniques also revealed that microbial biomass was collected on 0.22 μm pore-size GV microbial eukaryote assemblages displayed seasonality Sterivex cartridges (Merck-Millipore) and stored at –80 °C patterns in surface marine waters [27, 28], but interestingly until nucleic acid extraction. not always in the deeper ocean [28]. Reports on the sea- For cytometry, unfiltered seawater samples were fixed at sonality of archaea and eukaryotes are scarce, but there are a final concentration of 1% glutaraldehyde, incubated for even fewer time series studies covering simultaneously the 15 min at ambient temperature in the dark, frozen in liquid three domains of life. Steele et al. [29] identified the nitrogen and stored at −80 °C. Cytometry analyses were microorganisms that co-occurred during a 3-year study at performed on a Becton Dickinson FacsCalibur. Cells were the SPOT station (Southern California, USA). At the same excited at 488 nm and discriminated by SSC and red site, a 21-day study of the dynamics of phytoplankton, fluorescence (measured at 670 nm; chlorophyll content). archaea and bacteria revealed a rapid succession of micro- Orange fluorescence (measured at 585 ± 21 nm), produced bial species during a bloom [30], which highlighted the by phycoerythrin, was used to discriminate Synechococcus importance of taking into account microbial interactions from Prochloroccocus populations [15]. when studying the seasonality of marine microbial com- The physicochemical (temperature, salinity, nitrite, munities. However, long-term surveys of the annual nitrate, ammonium, phosphate and silicate) and biological dynamics and succession of photosynthetic picoeukaryotes, (chlorophyll a) parameters were provided by the Service bacteria and archaea are currently lacking. Moreover, most d’Observation en Milieu Littoral (SOMLIT). time series have covered open ocean sampling sites and there are very few studies dealing with the long term DNA extraction, amplification and sequencing monitoring of microbial communities at coastal sites. In the Mediterranean Sea, coastal environments are characterized The nucleic acid extraction followed protocols published by quite variable conditions caused by land to sea transfer earlier [25]. Briefly, the Sterivex filters were thawed on ice, of nutrients, organic matter and pollutants through seasonal followed by addition of lysis buffer (40 nM EDTA, 50 nM river discharge during periods of strong precipitations. In Tris, 0.75 M sucrose) and 25 µl of lysozyme (20 mg ml−1). such fluctuating environments, predictable patterns of The filters were then incubated on a rotary mixer at 37 °C reoccurring microbial communities would be less likely. for 45 min. The 8 µl of Proteinase K (20 mg ml−1) and 26 µl The main objective of this study was to test if the of sodium dodecyl sulfate (20% v/v) were added before eukaryotic phytoplankton, bacteria and archaea commu- incubating at 55 °C for 1 h. Total DNA was extracted and nities demonstrated significant patterns of rhythmicity at a purified with the Qiagen AllPrep kit (Qiagen, Hilden, coastal site. We conducted a 7-year survey of the taxonomic Germany) following the kit’s protocol. diversity of microbial plankton community at the Banyuls Specific primer pairs were used to target different Bay microbial observatory (SOLA) in the North Western domains of life. We used primers 515 F (5’-GTGY Mediterranean Sea, and investigated the environmental CAGCMGCCGCGGTA) [31] and NSR951 (5’-TTG Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

GYRAATGCTTTCGC) [32] to amplify the V4 region of For some ASVs, in order to obtain a finer taxonomical 18S rRNA eukaryote gene. Primers 27 F (5’-AGRGTTY resolution, we did an additional BLAST [41] search (blastn, GATYMTGGCTCAG) [33] and 519 R (5’-GTNTTAC 95% minimum similarity), which results can be found in the NGCGGCKGCTG) [34] were used for regions V1-V3 of column “Blast” of the supplementary table 1. We also did a the bacterial 16S rRNA gene, and finally primers 519 F PR2 [42] assignation for the rhythmic eukaryotic phyto- (5’-CAGCMGCCGCGGTAA) [35] and 1041 R (5’-GG plankton (supplementary table 1). In this study, we aimed to CCATGCACCWCCTCTC) [36] to amplify regions V4-V6 focus more specifically on autotrophic picoeukaryotes in of the archaeal 16S rRNA gene. order to highlight the co-occurrence patterns and rhythmi- As with all primers, there can biases introduced during city of phototrophs versus heterotrophs. We have therefore the amplification steps, either because some taxa can be selected a subset of the eukaryotic datasets by retaining preferentially amplified, or because of the uneven number sequences belonging to the divisions: Chlorophyta, Dino- of rRNA gene copies between taxa. A known example is the flagellata (without including Syndiniales, which are para- absence of haptophytes when classical 18S rRNA V4 pri- sitic), Ochrophyta and Haptophyta. Here we considered all mers are used [37]. Our eukaryote primers do amplify non-parasitic Dinoflagellata to be photosynthetic, but it haptophytes, but no primers are perfect, we hope to have should be noted that organisms from this group display a reduced primer biases in this study. range of metabolisms: phototrophic, mixotrophic and het- Sequencing was carried out with Illumina MiSeq 2 × 300 erotrophic [43]. bp kits by Research and Testing Laboratory (Lubbock, Texas). We noticed that the R2 reads were of lower quality Statistics and therefore chose to conduct our analysis with R1 reads only (300 bp). Having a good quality R2 reads would have The Lomb Scargle periodogram (LSP) was used to deter- been more informative. It could have improved taxa dif- mine if periodic patterns were present in microbial ASVs. ferentiation, taxonomic assignation and overall sequence The LSP, based on the Fourier transform, was originally quality. However, we remain confident, considering the adapted by astrophysicists to detect periodic signals in time length of the R1, that our data are robust. All the sequences series that were unevenly sampled due to limited access to were deposited in NCBI under accession number telescopes and varying weather conditions [44, 45]. The SRP139203. LSP was then successfully used in biological studies to determine the periodicity of an unevenly sampled signal Sequence analysis [46]. Owing to the robustness of the method and the fact that the sampling effort at SOLA was unevenly spaced, the The analysis of the raw sequences was done by following LSP appeared as the best tool for our study. Computing the the standard pipeline of the DADA2 package (https:// peak normalized power (PNmax) of the LSP was accom- benjjneb.github.io/dada2/index.html, version 1.6) in “R” plished via the “Lomb” package (https://cran.r-project. (https://cran.r-project.org) with the following parameters: org/web/packages/lomb/) in the “R” software. ASVs were trimLeft = 21, maxN = 0, maxEE = c(5,5), truncQ = 2. considered rhythmic when they had a PNmax > 10. The Briefly, the package includes the following steps: filtering, threshold for PNmax is automatically calculated by the dereplication, sample inference, chimera identification, and package. In summary, the LSP gives both the significance merging of paired-end reads [38]. DADA2 infers exact of the rhythmicity and the period of the rhythm. The LSP amplicon sequence variants (ASVs) from sequencing data, looks for all possible rhythmic patterns in a signal, instead of building operational taxonomic units from regardless of their period. To estimate the time of the year sequence similarity. In total, we had 159, 160 and of maximal abundance, we determined for each year and 158 samples for the eukaryotic phytoplankton, bacteria and each rhythmic ASV the week of the year with the highest archaea datasets respectively, and an average of ca. 27,000, number of sequences. Then we selected, over the entire time 29,000 and 16,000 reads per sample respectively. The series, the week that most often showed highest number of sequence data were normalized by dividing counts by sequences. sample size. This could influence our seasonality analyses, The Shannon index, to estimate community diversity, but considering our raw data, we found that the most was calculated for each sample and for eukaryotic phyto- appropriate transformation was to use proportional abun- plankton, bacteria and archaea, respectively, with the dances [39]. The taxonomy assignments were done with the function “diversity” from the “Vegan” package in “R” SILVA v.128 database (https://www.arb-silva.de/ (https://cran.r-project.org/web/packages/vegan/). documentation/release-128/) and the “assignTaxonomy” Distances between samples were calculated for eukar- function in DADA2 that implements the RDP naive yotic phytoplankton, bacteria and archaea based on com- Bayesian classifier method described in Wang et al. [40]. munity composition with a canonical correspondence S. Lambert et al. analyses (CCA). Contribution of environmental factors values at the beginning and the end of winter, and lower were added as arrows, and their significance was tested with values during late summer (Supplementary Fig. 1). Bac- an analysis of variance (ANOVA) from the “Vegan” terial communities had, on average, the highest diversity, package in "R". followed by eukaryotic phytoplankton and then archaeal Patterns of co-occurrences between taxa were measured communities. with the sparse partial least squares (sPLS) regression [47]. Canonical correspondence analyses (CCA) were per- The sPLS was used to relate the abundance matrices of formed on the eukaryotic phytoplankton, bacteria and eukaryotic phytoplankton against bacteria and archaea with archaea datasets to investigate the relationships between these parameters: ncomp = 3, mode = ‘regression’, in the community composition and measured environmental “mixOmics” package (https://cran.r-project.org/web/packa variables (Fig. 2a–c). The communities showed a strong ges/mixOmics/) in "R". Relationships between taxa were seasonal pattern but the environmental parameters that we then visualized by a heatmap with the “CIM” function, from measured explained only 7, 12 and 14% of the variance for the same package. the eukaryotic phytoplankton, bacteria and archaea com- Eukaryotic phytoplankton, bacteria and archaea ASV munities respectively (Supplementary Table 2). The main tables containing reference sequences, taxonomy and pro- explaining factors were temperature (T), day length (DL) portional abundance in the different samples are available as for the three datasets, and also Nitrate (NO3) and Salinity supplementary table 1. (S) for bacteria (ANOVA, p = 0.001). Temperature and day length explained close to half of the total variance for eukaryotic phytoplankton, bacteria and archaea (Supple- Results mentary Table 2). The eukaryotic phytoplankton communities grouped together according to the month of sampling. Environmental conditions The communities showed more divergence on the CCA plots during the months of April and May, whereas they Chlorophyll a concentrations showed yearly reoccurring were grouped during the other months (Fig. 2a). Bacterial patterns with maxima reaching up to 2.50 µg l−1 during the communities showed a similar seasonal structure with winter to spring transitions, and minima at 0.04 µg l−1 higher separation between samples from March to June during summer months (Fig. 1). Similarly, temperature (Fig. 2b). Finally, the archaea had a comparable structure of levels showed yearly patterns but with much less pro- monthly successions, but with highest variability between nounced inter-annual variations. Water temperature at samples from July to October (Fig. 2c). SOLA were warmest during the months of August and From 2007 to 2015, at the division level, the photo- September usually, reaching 22 °C, and coldest between synthetic picoeukaryote community was composed of February and March, with values as low as 10 °C. Salinity Dinoflagellata, Chlorophyta, Ochrophyta and Haptophyta fluctuated from 38.49 to 34.27 psu, with an average of (44.01% of the sequences, 29.45, 13.23, 13.31% respec- 37.63 psu. Nitrate levels extended from undetectable to tively) (Supplementary Fig. 2). Dinoflagellata were domi- 9.52 µmol l−1 with an average of 0.90 µmol l −1. Phosphate nated by Dinophyceae (99.19% of the sequences) and concentrations varied from 0.01 µmol l−1 to 0.36 µmol l −1 Chlorophyta by Mamiellophyceae (94.36%). Within with an average of 0.04 µmol l–1. Nitrate, phosphate and Mamiellophyceae, three main genera were found, Micro- chlorophyll a concentrations had highest values at the monas, Bathycoccus and Ostreococcus (64.59, 31.89 and winter/spring transition and lowest in summer. However, 3.49%, respectively) (Supplementary Fig. 2). Bacteria salinity, nitrate and phosphate concentrations varied more (Supplementary Fig. 3) were dominated by the phyla Pro- than average in November 2011, March 2013 and January teobacteria (76.74%) and Cyanobacteria (12.12%). The 2014 when decreases in salinity levels co-occurred with main contributors of the Proteobacteria were Alphapro- increases in nitrate and phosphate levels (Fig. 1). teobacteria (89.79%, mainly SAR11) and Gammaproteo- bacteria (9.93%). Synechococcus ASVs represented 95.8% Eukaryotic phytoplankton, bacteria and archaea of Cyanobacteria sequences. Finally, archaea were divided community composition between the Thaumarchaeota (64.36%) and the Eur- yarchaeota (35.07%) (Supplementary Fig. 4). Overall, the datasets yielded 6398, 6242 and 918 ASVs for the eukaryotic phytoplankton, bacterial and archaeal com- Rhythmicity of the environmental and biological munities respectively. Within the eukaryotes, 1801 ASVs compartments corresponded to autotrophs (eukaryotic phytoplankton). The Shannon index showed similar patterns of diversity for In order to test if environmental factors and microbial taxa autotrophic eukaryotes, bacteria and archaea, with higher had significant rhythmic patterns during the 7-year time Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

Fig. 1 Salinity, nitrates (NO3), phosphates (PO4), chlorophyll a (CHLA) and temperature from 2008 to 2015 at the SOLA station in the Banyuls Bay series, the Lomb Scargle periodogram (LSP) algorithm was rhythmicity. The rhythmic ASVs and environmental factors applied to the eukaryotic phytoplankton, bacteria, archaea all had a period of one year. Theses rhythmic microbial and environmental datasets. The most rhythmic environ- ASVs were selected for further detailed analysis. mental parameters were day length and temperature with a PNmax score of 60.00 and 55.67 respectively. Other Timing of yearly reoccurrences and relative rhythmic factors were NO2,NO3, chlorophyll a and abundance of rhythmic ASVs NH4 but with lower PNmax scores of 37.17, 24.27, 21.37 and 13.44, respectively. SIOH4,PO4 and salinity had Among the 135 ASVs (Fig. 3, Supplementary Table 3) that PNmax scores that did not cross the statistical threshold to showed signiﬁcant reoccurrences throughout the year, dif- be considered rhythmic (PNmax scores of 9.95, 7.03 and ferent domains displayed different patterns. Bacterial 5.45, respectively). A total of 15 picoeukaryote, 89 bacteria rhythmic ASVs showed phases of maximal abundance that and 31 archaea ASVs had signiﬁcant patterns of spread throughout the year, whereas eukaryotic S. Lambert et al.

Fig. 2 Canonical correspondence analyses (CCA) of the eukaryotic phytoplankton (a), bacteria (b), and archaea (c) community composition in relation to environmental factors. The communities are color coded according to the month of sampling. The arrows represent the different environmental factors (T: temperature, DL: day length, NH4: ammonium, NO3: nitrates, NO2: nitrites, PO4: phosphates, SIOH4: silicates, S: salinity)

November to April, while archaeal rhythmic ASVs had maximal abundance from September to March. On average 30.5% of the eukaryotic phytoplankton sequences were rhythmic but the proportion varied throughout the year. Rhythmic ASVs represented up to 96% of the sequences in January and as low as 2.5% of the sequences in July (Fig. 4b). All classes followed a similar pattern with high levels (50 to 60% of total sequences) from mid-Autumn to mid-Spring (October to April) and lower levels (less than 15% of total sequences) during the rest of the year. The lowest number of rhythmic sequences were seen during the summer months (Fig. 4b). Flow cytometry showed that picoeukaryotes had low abundances during the summer months and high abundances during winter months (Fig. 5). At the eukaryotic class level, (Fig. 4a), the Mamiello- phyceae rhythmic ASVs were found mostly from the end of November to the end of March. The Dinophyceae rhythmic ASVs had peaks of abundances year-round. The Dictyo- chophyceae rhythmic ASV was only abundant at the beginning of February. Within Mamiellophyceae, the Bathycoccus prasinos ASV peaked around the middle of February (7th week of the year) (Fig. 4c) with a distribution going from January to April (Fig. 4d). Micromonas com- moda was recurrent from December to the end of March (Fig. 4c) and distributed from February to April (Fig. 4d). Micromonas sp.1 ASV was more present at the end of November (Fig. 4c) with a distribution from November to February (Fig. 4d). Micromonas bravo, however, had ASVs peaks from December to February (Fig. 4c) and was present from October to April (Fig. 4d). Rhythmic bacterial ASVs were present throughout the year (Fig. 6a), and represented in average 31.3% of the sequences, with variations from 18 to 45.7% of the sequences (Fig. 6b). The contributors to the rhythmic ASVs were Acidimicrobiia, Alphaproteobacteria, Betaproteo- bacteria, Cyanobacteria, Flavobacteria, Gammaproteo- bacteria, SAR202 and candidate Proteobacteria SPOTSOCT00m83 (Fig. 6a, b). The different rhythmic bacterial classes showed different types of patterns. The Acidimicrobiia, Gammaproteo- bacteria, SAR202 and candidate Proteobacteria SPOTSOCT00m83 showed high numbers from October to phytoplankton and archaeal rhythmic ASVs phases were April and were almost absent during the summer months conﬁned to certain moments of the year. Eukaryotic phy- (Fig. 6b). They displayed similar reoccurrence patterns as toplankton rhythmic ASVs had maximal abundance from well, mainly from December to February (Fig. 6a). Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

Fig. 3 Polar plot showing when during the year the rhythmic ASVs rhythmicity (PNmax = 10). The ASVs are color coded according to reoccur and the strength of reoccurrence (PNmax, calculated via the which domain of life they belong to LSP). The black circle shows the statistical threshold for signiﬁcant

Cyanobacteria rhythmic ASVs demonstrated an opposite during winter months (Supplementary Fig. 5B), except for pattern, with high levels during the warm summer and SAR11IIIa which had higher sequence abundance from autumn months (March to October) and low levels the rest June to November (Supplementary Fig. 5B). of the year (Fig. 6b). Cytometry data showed the same Finally, archaeal rhythmic ASVs had maximum occur- seasonal pattern in terms of cell abundance (Fig. 5). Fla- rences from the end of August to March, both for Eur- vobacteria had similar patterns as cyanobacteria. However, yarchaeota and Thaumarchaeota (Fig. 6c). Rhythmic ASVs their reoccurrence patterns were different. Rhythmic Cya- dominated the dataset as they represented an average of nobacteria ASVs reoccurred from the end of March to 74.6% of total sequence numbers, ranging from 47.3 to October, whereas Flavobacteria ASVs had two periods of 89.2% (Fig. 6d). Within the Euryarchaeota phylum, maximal reoccurrence, one from March to July and another rhythmic ASVs of Marine group II (MGII) and Marine during December (Fig. 6a). Betaproteobacteria ASVs were group III (MGIII) were found. Rhythmic MGII ASVs more abundant from January to May and were absent the showed reoccurrence patterns from September to March rest of the year (Fig. 6b), and were only recurrent at the end (Fig. 6c) and highest relative sequence numbers from July of February (Fig. 6a). Alphaproteobacteria rhythmic ASVs to October (Fig. 6d). MGIII rhythmic ASVs had a more displayed similar sequence numbers throughout the year, restrained occurrence, from end of November to beginning accounting for half of the rhythmic ASVs sequence num- of December (Fig. 6c) and were less present in relative bers (15%) (Fig. 6b). Similarly, the Alphaproteobacteria abundance (Fig. 6d). The Thaumarchaeota rhythmic ASVs ASVs reoccurrences covered the whole year except for displayed high levels of presence throughout the year with March (Fig. 6a). the exception of the months of September. The months Amongst the rhythmic Alphaproteobacteria, a majority preceding and succeeding September showed a steady of ASVs belonged to SAR11. All sub-groups of SAR11 decrease and increase of relative sequence number, (SAR11Ia, SAR11Ib, SAR11Ic, SAR11IIa, SAR11IIIa and respectively (Fig. 6d). Thaumarchaeota had high occur- SAR11IV) had high numbers of rhythmic ASVs from rences all year, except from March to May (Fig. 6c). September to the end of February (Supplementary Fig. 5A). We also observed a large number of ASVs that were not These groups also showed higher number of sequences rhythmic and thus had peaks of abundance at different S. Lambert et al.

Fig. 4 Polar plots representing the rhythmic eukaryotic phytoplankton averaged per week of the year for eukaryotic phytoplankton classes (b) classes (a) and the rhythmic Mamiellophyceae ASVs (c). The bar plots and Mamiellophyceae ASVs (d) show the proportion of sequences belonging to rhythmic ASVs moments from year to year. Non-rhythmic ASVs had dif- ASV 00112 and ASV 00266). A Dinophyceae (ASV ferent patterns of seasonal dynamics. Some ASVs, like the 00011) also had a high correlation (0.55) with the same Gymnodiniphycidae ASV00020, were absent from most of Alphaproteobacteria ASVs. Other high correlations were the samples but shows sudden and irregular peaks of found between Bathycoccus prasinos and Alpha- and abundance (Supplementary Fig. 6). Other, like the Gym- Gammaproteobacteria. Micromonas bravo (ASV 00002) nodiniphycidae ASV00036, were more frequent and had also had high correlations with an Alphaproetobacteria irregular peaks of sequence abundance that co-occurred (ASV 00112). A Dinophyceae ASV (ASV00053) displayed with irregular environmental events such as freshening sea a speciﬁc high correlation with a group of bacteria that were surface waters and increased nitrate concentrations (Sup- not correlated to other eukaryotic phytoplankton. This is plementary Fig. 6). probably due to the fact that Dinophyceae is the only rhythmic picoeukaryote to peak in summer (Fig. 4a). Co-occurrence at the ASV level The archaea vs. picoeukaryote heatmap revealed high correlation ( > 0.5) between Bathycoccus prasinos To determine co-occurrences, heatmaps were created with and MGII ASVs (ASV 00050 and ASV 00008). Micro- the rhythmic ASVs after calculating Sparse Partial Least monas bravo (ASV 00040) showed a similar trend. On the Squares (sPLS) regressions for one dataset against the other other hand, Micromonas commada (ASV 00084) had high (bacteria vs. picoeukaryote, bacteria vs. archaea and archaea correlations ( > 0.5) with MGIII ASVs (ASV 00012 vs. picoeukaryote). For bacteria vs. picoeukaryotes and ASV 00028). As with the bacteria dataset, the Dino- (Fig. 7a), the highest correlation scores (>0.6) were between phyceae, ASV 00053, displayed high correlations when all Micromonas sp.1 (ASV 00013) and a SAR11 sequence other eukaryotic phytoplankton ASVs had low correlations (ASV 00054) as well as 3 Rhodospirillaceae (ASV 00020, (Fig. 7b). Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

Fig. 5 Photosynthetic picoeukaryote and cyanobacteria abundance determined by ﬂow cytometry from 2009 to 2015 at the SOLA station in the Banyuls Bay

In the bacteria vs. archaea heatmap the main co- Mediterranean. It has a marked seasonality but interestingly occurrences were observed between a Thaumarchaeota it is also characterized by strong and ephemeral inputs of ASV and a Gammaproteobacteria ASV, as well as between nutrients brought from sediment mixing during episodic a MGII and Alphaproteobacteria ASV (Supplementary winter storms and during flash floods from incoming rivers Fig. 7). [49]. Nutrients are known to strongly structure communities by promoting planktonic blooms and by stimulating the growth of certain microbes [12, 50]. However, despite Discussion irregular nutrient supply from year to year, as illustrated by salinity and phosphate variations during winter and spring Our 7-year survey in the NW Mediterranean Sea showed (Fig. 1), we could still observe a large number of rhythmic that within all domains of life some taxa showed significant eukaryotic phytoplankton, bacteria and archaea sequences. patterns of rhythmicity with a one year period. The number The CCA analysis (Fig. 2) confirmed that day length and of rhythmic taxa differed between domains. Phototrophic temperature were major factors structuring the communities picoeukaryotes had 1% of rhythmic ASVs, bacteria 3.1 % and we can suppose that they directly or indirectly control and archaea 3.4%, but these ASVs represented a large the dynamics of the rhythmic taxa. proportion of the total number of sequences (31.3, 31.6 and Day length has been shown to be a strong driver of 75.5%, respectively). The large proportion of rhythmic community structure in temperate and polar marine envir- sequences supports the idea of microbial communities that onments such as the English Channel [14], or a high-Arctic come back year after year at the same season. The concept fjord [50]. Temperature is another strong driver as it can of re-occurring communities has been demonstrated in affect gene expression and subsequently the structure and several long term studies [14, 20, 24] but coastal observa- the function of the microbial communities [51]. The avail- tions are quite scarce [48]. The Banyuls Bay is a coastal site ability of nutrients has also been shown to be an important with seasonal characteristics specific to the NW factor in community composition as demonstrated in the S. Lambert et al.

Fig. 6 Polar plots representing the rhythmic bacteria ASVs (a) and the rhythmic archaea ASVs (c). The bar plots show the proportion of sequences belonging to rhythmic ASVs averaged per week of the year, for bacteria (b) and archaea (d)

BATS time series in the Atlantic Ocean [3]. Our data from this order. In diatoms from northern Norwegian coastal the Banyuls Bay shows that in a coastal ecosystem, envir- waters, it has been reported that the timing of the spring onmental parameters like temperature and day length have bloom varies little from year to year whether water stra- such a structuring effect that sporadic meteorological events tification had occurred or not [53]. The authors hypothe- do not appear to impact the overall microbial rhythms of re- sized that the photoperiod was the major factor that occurring dominant groups of eukaryotic phytoplankton, relieved diatoms resting spores from dormancy, leading to bacteria and archaea. However, even though rhythmic seasonal blooms. However, the internal mechanisms ASVs could be predominately influenced by day length and triggering these rhythms remain unknown since the pre- temperature, we observed non-rhythmic ASVs, which were sence of circadian clocks remain to be shown in diatoms. influenced by irregular environmental factors. For example, Amongst the prokaryotes, cyanobacteria are the only the dynamics of the Gymnodiniphycidae ASV00036 was known group to have a genuine circadian clock [7]andthe associated to the irregular peaks of nitrate concentration occurrence of circadian clock remain to be established in (Supplementary Fig. 6). heterotrophic bacteria and archaea. The rhythmicity of The importance of day length in driving the rhythm of some heterotrophic microorganisms could thus be gov- individual microorganisms brings the question whether erned directly by day length or indirectly through inter- seasonality is driven by circadian clocks in marine actions with the rhythmic autotrophs. Interestingly, microbes. The presence of a functional circadian clock altogether, eukaryotic and prokaryotic autotrophs were governing day/night biological processes bas been present during the entire year, but they showed clear demonstrated in the mamiellophyceae Ostreococcus [8, differences in their seasonal dynamics. Picoeukaryotes 52], however, the existence of a photoperiod dependent had highest abundance from autumn to spring, and cya- regulation of blooms remains to be established formally in nobacteria during the summer. Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

Fig. 7 Heatmap based on a sPLS regression showing co-occurrences (a) and between rhythmic eukaryotic phytoplankton ASVs and archaea between rhythmic eukaryotic phytoplankton ASVs and bacteria ASVs ASVs (b). Correlations > 0.4 are highlighted S. Lambert et al.

We observed a large number of rhythmic ASV sequen- phytoplankton and prokaryotes ASVs. The most significant ces, which were mainly seen within abundant members of correlations were found between Mamiellophyceae and the the communities. The eukaryotic phytoplankton were alphaproteobacteria SAR11. This co-occurrence could be represented primarily by Mamiellophyceae and Dinophy- explained by the fact that Micromonas and SAR11 might ceae, which have important ecological roles as primary interact by exchanging compounds such as vitamins, producers and as links in the predator/prey food chain [54]. growth factors and organic carbon [58, 59]. However, Among prokaryotic rhythmic ASVs, there were many SAR11 was recently shown to be auxotrophic to the thia- representative of the SAR11, known for being the most mine precursor 4-amino-5-hydroxymethyl-2-methyl pyr- common group of marine bacteria [23]. Seasonality has imidine [60], thus resulting in similar needs as Micromonas been observed for SAR11 and Flavobacteria groups earlier for thiamin precursors [61]. The co-occurrence of these two [15, 55]. Rhythmic archaea were found in MGII, MGIII and microbes therefore may be explained by their requirement Thaumarchaeota, which have already been shown to have for similar nutritional niches rather than by a relationship of reoccurring yearly patterns [20, 25]. The dominance of interdependency depending on environmental factors. abundant groups within rhythmic ASVs raises the question In conclusion, through the analysis of our time series we as whether our analysis underestimated the rhythmicity of demonstrated that a large proportion of members of less abundant ASVs. It should be noted, however, that some eukaryotic phytoplankton, bacteria and archaea datasets, rare ASVs, with occurrences of 0.034, 0.009 and 0.115% showed rhythmicity with a one year period of reoccurrence respectively for the eukaryotic phytoplankton, bacteria and over then entire time series. The main drivers of seasonality archaea dataset, were also found to be rhythmic. In agree- were photoperiod and temperature. Sporadic meteorological ment with our observations, Alonso-Sáez et al. [55] recently events and irregular nutrient supply characteristic of our showed that, also in a coastal system, both rare and abun- coastal site did not affect significantly the seasonality, dant bacterial species had patterns of rhythmicity in the indicating that the yearly rhythms were robust. Rhythmicity Atlantic Ocean [55] and that many species that remained was found in both autotrophs (picoeukaryotes and cyano- rare all year long also showed significant patterns of bacteria) and heterotrophic prokaryotes. Seasonal auto- rhythmicity. Rhythmicity of marine microbes, at the ASV trophs, which respond to light, may be setting the pace for level, remains to be verified in other sites as there have been rhythmic heterotrophs but similar environmental niches only few studies conducted at this level of resolution. While may be driving seasonality as well. the re-occurrence of entire communities is now well docu- mented [13, 14, 24], the long-term monitoring of individual Acknowledgements We are grateful to the captain and the crew of the ‘ ’ taxa is not common [12, 21] and the use of statistics to test RV Nereis II for their help in acquiring the samples. We thank the “Service d’Observation”, particularly Eric Maria and Paul Labatut, for patterns of ASV rhythmicity is even less frequent. their help in obtaining and processing of the samples. MT was sup- There have been very few studies looking at the temporal ported by a PhD fellowship from the Sorbonne Université and the dynamics across the three domains of life in marine Région Bretagne. We would like to thank the ABIMS platform in microbial ecosystems. One of the first long term study Roscoff for access to bioinformatics resources. This work was sup- ported by the French Agence Nationale de la Recherche through the covering the three domains did not focus on the rhythm of projects Photo-Phyto (ANR-14-CE02-0018) to FYB, and EUREKA the individual taxa but rather looked at co-occurrence net- (ANR-14-CE02-0004-01) to PEG. works [29]. They showed that correlations between microbes were more prevalent than correlations between Compliance with ethical standards microbes and environmental factors. This is probably due to the stability of the deep chlorophyll maximum at their study Conflict of interest The authors declare that they have no conflict of site [29]. More recently, Needham and Fuhrman looked at interest. the succession of phytoplankton, archaea and bacteria, but only during 6 months [30]. Another study in the same References ecosystem, relying on automated sampling, showed daily and highly dynamic population variations in the three 1. Antle MC, Silver R. Circadian insights into motivated behavior. domains of life, and extensively described the biological In: Simpson EH, Balsam PD, (eds). Behavioral neuroscience of motivation. Cham: Springer International Publishing; 2015. p. interactions that took place during the sampling period [56]. 137–69. A study looking at bacterioplankton diversity and phyto- 2. MacDonald CC, McMahon KW. The flowers that bloom in the plankton microscopy counts, has shown that despite inter- spring: RNA processing and seasonal flowering. Cell. – annual variations in phytoplankton blooms, bacter- 2003;113:671 72. 3. Treusch AH, Demir-Hilton E, Vergin KL, Worden AZ, Carlson ioplankton microdiversity patterns seem stable in both CA, Donatz MG, et al. Phytoplankton distribution patterns in the bloom and non-bloom conditions [57]. The present dataset northwestern Sargasso Sea revealed by small subunit rRNA genes showed high co-occurrence between some eukaryotic from plastids. ISME J. 2012;6:481–92. Rhythmicity of coastal marine picoeukaryotes, bacteria and archaea despite irregular environmental. . .

4. Sverdrup HU. On vernal blooming of phytoplankton. J Cons Exp the Bermuda Atlantic Time-series study site by phylogenetic Mer. 1953;18:287–95. placement of pyrosequences. ISME J. 2013;7:1322–32. 5. Behrenfeld MJ. Abandoning Sverdrup’s critical depth hypothesis 24. Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV, on phytoplankton blooms. Ecology. 2010;91:977–989. Naeem S. Annually reoccurring bacterial communities are pre- 6. Fowler S, Lee K, Onouchi H, Samach A, Richardson K, Morris B, dictable from ocean conditions. Proc Natl Acad Sci USA. et al. GIGANTEA: a circadian clock-controlled gene that reg- 2006;103:13104–09. ulates photoperiodic flowering in Arabidopsis and encodes a 25. Hugoni M, Taib N, Debroas D, Domaizon I, Jouan Dufournel I, protein with several possible membrane-spanning domains. Bronner G, et al. Structure of the rare archaeal biosphere and EMBO J. 1999;18:4679–88. seasonal dynamics of active ecotypes in surface coastal waters. 7. Cohen SE, Golden SS. Circadian rhythms in Cyanobacteria. Proc Natl Acad Sci USA. 2013;110:6004–9. Microbiol Mol Biol Rev. 2015;79:373–85. 26. Winder M, Cloern JE. The annual cycles of phytoplankton bio- 8. Corellou F, Schwartz C, Motta J-P, Djouani-Tahri EB, Sanchez F, mass. Philos Trans R Soc B: Biol Sci. 2010;365:3215–26. Bouget F-Y. Clocks in the green lineage: comparative functional 27. Brannock PM, Ortmann AC, Moss AG, Halanych KM. Meta- analysis of the circadian architecture of the picoeukaryote barcoding reveals environmental factors influencing spatio‐temporal ostreococcus. Plant Cell. 2009;21:3436–49. variation in pelagic micro‐eukaryotes. Mol Ecol. 2016;25:3593–604. 9. Edmunds L, Adams K. Clocked cell cycle clocks. Science. 28. Kim DY, Countway PD, Jones AC, Schnetzer A, Yamashita W, 1981;211:1002–13. Tung C, et al. Monthly to interannual variability of microbial 10. Jacquet S, Partensky F, Lennon J-F, Vaulot D. Diel patterns of eukaryote assemblages at four depths in the eastern North Pacific. growth and division in marine picoplankton in culture. J Phycol. ISME J. 2014;8:515–30. 2001;37:357–69. 29. Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, 11. Alonso-Sáez L, Balagué V, Sá EL, Sánchez O, González JM, et al. Marine bacterial, archaeal and protistan association networks Pinhassi J, et al. Seasonality in bacterial diversity in north-west reveal ecological linkages. ISME J. 2011;5:1414–25. Mediterranean coastal waters: assessment through clone libraries, 30. Needham DM, Fuhrman JA. Pronounced daily succession of fingerprinting and FISH: Seasonality in marine bacterial diversity. phytoplankton, archaea and bacteria following a spring bloom. FEMS Microbiol Ecol. 2007;60:98–112. Nat Microbiol. 2016;1:16005. 12. Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, 31. Parada AE, Needham DM, Fuhrman JA. Every base matters: Patel A, et al. Temporal variability and coherence of euphotic assessing small subunit rRNA primers for marine microbiomes zone bacterial communities over a decade in the Southern Cali- with mock communities, time series and global field samples. fornia Bight. ISME J. 2013;7:2259–73. Environ Microbiol. 2016;18:1403–14. 13. Cram JA, Chow C-ET, Sachdeva R, Needham DM, Parada AE, 32. Mangot J-F, Domaizon I, Taib N, Marouni N, Duffaud E, Bronner Steele JA, et al. Seasonal and interannual variability of the marine G, et al. Short-term dynamics of diversity patterns: evidence of bacterioplankton community throughout the water column over continual reassembly within lacustrine small eukaryotes: short- ten years. ISME J. 2015;9:563–80. term dynamics of small eukaryotes. Environ Microbiol. 14. Gilbert JA, Steele JA, Caporaso JG, Steinbrück L, Reeder J, 2013;15:1745–58. Temperton B, et al. Defining seasonal marine microbial commu- 33. Lane D J. 16S/23S rRNA sequencing. In: Stackebrandt E, nity dynamics. ISME J. 2012;6:298–308. Goodfellow M, editors. Nucleic acid techniques in bacterial sys- 15. Salter I, Galand PE, Fagervold SK, Lebaron P, Obernosterer I, tematics. Chichester, United Kingdom: John Wiley and Sons; Oliver MJ, et al. Seasonal dynamics of active SAR11 ecotypes in 1991. pp. 115–175. the oligotrophic Northwest Mediterranean Sea. ISME J. 34. Turner S, Pryer KM, Miao VPW, Palmer JD. Investigating deep 2015;9:347–60. phylogenetic relationships among cyanobacteria and plastids by 16. Tonkin JD, Bogan MT, Bonada N, Rios-Touma B, Lytle DA. small subunit rRNA sequence analysis. J Eukaryot Microbiol. Seasonality and predictability shape temporal species diversity. 1999;46:327–38. Ecology. 2017;98:1201–16. 35. Ovreås L, Forney L, Daae FL, Torsvik V. Distribution of bac- 17. Ladau J, Sharpton TJ, Finucane MM, Jospin G, Kembel SW, terioplankton in meromictic Lake Saelenvannet, as determined by O’Dwyer J, et al. Global marine bacterial diversity peaks at high denaturing gradient gel electrophoresis of PCR-amplified gene latitudes in winter. Isme J. 2013;7:1669. fragments coding for 16S rRNA. Appl Environ Microbiol. 18. Bunse C, Pinhassi J. Marine bacterioplankton seasonal succession 1997;63:3367–73. dynamics. Trends Microbiol. 2017;25:494–505. 36. Kolganova TV, Kuznetsov BB, Tourova TP. Designing and 19. Fuhrman JA, Cram JA, Needham DM. Marine microbial com- testing oligonucleotide primers for amplification and sequencing munity dynamics and their ecological interpretation. Nat Rev of archaeal 16S rRNA. Genes. 2002;71:4. Microbiol. 2015;13:133–146. 37. Liu H, Probert I, Uitz J, Claustre H, Aris-Brosou S, Frada M, et al. 20. Galand PE, Gutiérrez-Provecho C, Massana R, Gasol JM, Casa- Extreme diversity in noncalcifying haptophytes explains a major mayor EO. Inter-annual recurrence of archaeal assemblages in the pigment paradox in open oceans. Proc Natl Acad Sci USA. coastal NW Mediterranean Sea (Blanes Bay Microbial Observa- 2009;106:12803–08. tory). Limnol Oceanogr. 2010;55:2117–25. 38. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, 21. Lindh MV, Sjöstedt J, Andersson AF, Baltar F, Hugerth LW, Holmes SP. DADA2: High-resolution sample inference from Lundin D, et al. Disentangling seasonal bacterioplankton popu- Illumina amplicon data. Nat Methods. 2016;13:581–3. lation dynamics by high-frequency sampling: high-resolution 39. Weiss S, Xu ZZ, Peddada S, Amir A, Bittinger K, Gonzalez A, temporal dynamics of marine bacteria. Environ Microbiol. et al. Normalization and microbial differential abundance strate- 2015;17:2459–76. gies depend upon data characteristics. Microbiome 2017;5:27. 22. Treusch AH, Vergin KL, Finlay LA, Donatz MG, Burton RM, 40. Wang Q, Garrity GM, Tiedje JM, Cole JR. Naïve bayesian clas- Carlson CA, et al. Seasonality and vertical structure of microbial sifier for rapid assignment of rRNA sequences into the new bac- communities in an ocean gyre. ISME J. 2009;3:1148–63. terial taxonomy. Appl Environ Microbiol. 2007;73:5261–7. 23. Vergin KL, Beszteri B, Monier A, Cameron Thrash J, Temperton 41. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic B, Treusch AH, et al. High-resolution SAR11 ecotype dynamics at local alignment search tool. J Mol Biol. 1990;215:403–10. S. Lambert et al.

42. Guillou L, Bachar D, Audic S, Bass D, Berney C, Bittner L, et al. 52. Monnier A, Liverani S, Bouvet R, Jesson B, Smith JQ, Mosser J, The protist ribosomal reference database (PR2): a catalog of et al. Orchestrated transcription of biological processes in the unicellular eukaryote small sub-unit rRNA sequences with curated marine picoeukaryote ostreococcus exposed to light/dark cycles. taxonomy. Nucleic Acids Res. 2013;41:D597–604. BMC Genom. 2010;11:192. 43. Sanchez-Puerta MV, Lippmeier JC, Apt KE, Delwiche CF. 53. Eilertsen HC, Sandberg S, Tøllefsen H. Photoperiodic control of Plastid genes in a non-photosynthetic dinoﬂagellate. Protist. diatom spore growth: a theory to explain the onset of phytoplankton 2007;158:105–17. blooms. Mar Ecol Progress Ser Oldendorf. 1995;116:303–7. 44. Lomb NR. Least-squares frequency analysis of unequally spaced 54. Massana R. Eukaryotic picoplankton in surface o. Annu Rev data. Astrophys Space Sci. 1976;39:447–62. Microbiol. 2011;65:91–110. 45. Scargle JD. Studies in astronomical time series analysis. II- 55. Alonso-Sáez L, Díaz-Pérez L, Morán XAG. The hidden seasonality Statistical aspects of spectral analysis of unevenly spaced data. of the rare biosphere in coastal marine bacterioplankton: Seasonality Astrophys J. 1982;263:835–53. of the rare biosphere. Environ Microbiol. 2015;17:3766–80. 46. Ruf T. The lomb-scargle periodogram in biological rhythm 56. Needham DM, Fichot EB, Wang E, Berdjeb L, Cram JA, Fichot research: analysis of incomplete and unequally spaced time-series. CG, et al. Dynamics and interactions of highly resolved marine Biol Rhythm Res. 1999;30:178–201. plankton via automated high-frequency sampling. The ISME J. 47. Lê Cao K-A, Rossow D, Robert-Granié C, Besse P. Sparse PLS: 2018. https://doi.org/10.1038/s41396-018-0169-y. Variable selection when integrating omics data. 2008. 57. Chafee M, Fernàndez-Guerra A, Buttigieg PL, Gerdts G, Eren 48. Nelson JD, Boehme SE, Reimers CE, Sherrell RM, Kerkhof LJ. AM, Teeling H, et al. Recurrent patterns of microdiversity in a Temporal patterns of microbial community structure in the mid- temperate coastal marine environment. ISME J. 2018;12:237–52. atlantic bight: spatio-temporal variability of coastal marine bac- 58. Alonso-Saez L, Gasol JM. Seasonal variations in the contributions teria. FEMS Microbiol Ecol. 2008;65:484–93. of different bacterial groups to the uptake of low-molecular- 49. Charles F, Lantoine F, Brugel S, Chrétiennot-Dinet M-J, weight compounds in northwestern mediterranean coastal waters. Quiroga I, Rivière B. Seasonal survey of the Appl Environ Microbiol. 2007;73:3528–35. phytoplankton biomass, composition and production in a littoral 59. Paerl RW, Bouget F-Y, Lozano J-C, Vergé V, Schatt P, Allen EE, NW mediterranean szsite, with special emphasis on the pico- et al. Use of plankton-derived vitamin B1 precursors, especially planktonic contribution. Estuar, Coast Shelf Sci. 2005;65: thiazole-related precursor, by key marine picoeukaryotic phyto- 199–212. plankton. ISME J. 2017;11:753–65. 50. Marquardt M, Vader A, Stübner EI, Reigstad M, Gabrielsen TM. 60. Carini P, Campbell EO, Morré J, Sañudo-Wilhelmy SA, Cameron Strong seasonality of marine microbial eukaryotes in a high-arctic Thrash J, Bennett SE, et al. Discovery of a SAR11 growth fjord (Isfjorden, in West Spitsbergen, Norway). Appl Environ requirement for thiamin’s pyrimidine precursor and its distribution Microbiol. 2016;82:1868–80. in the Sargasso Sea. ISME J. 2014;8:1727–38. 51. Ward CS, Yung C-M, Davis KM, Blinebry SK, Williams TC, 61. Paerl RW, Bertrand EM, Allen AE, Palenik B, Azam F. Vitamin Johnson ZI, et al. Annual community patterns are driven by sea- B1 ecophysiology of marine picoeukaryotic algae: strain‐speciﬁc sonal switching between closely related marine bacteria. ISME J. differences and a new role for bacteria in vitamin cycling. Limnol 2017;11:1412–22. Oceanogr. 2015;60:215–28.